Research Of The Reactive Power Optimization Of Static And Dynamic Of Oil Field Distribution Network Based On Genetic Neural Network

At present, most of oil exploitations in China are in the mid-late period. The recovery ratio decreases while the costs increase. Owing to special operation mode of pumping unit and the diminishing oil content, the distribution network of the oil field has many shortcomings, such as high ratio of reactive content and network loss, low power factor and so on, which cause serious energy waste. So, it is important to research the method of reactive power optimization.Based on the analysis of the status quo of oilfield distribution network, the paper proposed an improved method, which based on genetic algorithm of neural network to carry out the static and dynamic reactive power optimization. It established the mathematic model of reactive power optimization, which took net loss ratio and voltage eligibility as the objective function. The electric parameters are calculated by using former back generation method to carry on trend analysis of nod network. A method, which combines the loss reduction at large and impedance moment of load power, is proposed in order to certain the specific location of static and dynamic nod of reactive compensation. In allusion to the up-down-stroke operation mode of pumping unit, the paper proposed a method, which combined Gray GM (1, 1) improved model and trend analysis, dynamically stimulate the distribution network. The result of stimulation provided the initial data for dynamic reactive power optimization.In static, dynamic reactive power optimization study, taking average load power as the object, utilized improved genetic algorithm to calculate the compensation capacity of static compensation nod. According to the load power average of 50%~150%, simulated parameters dynamicly to generate some new load power groups, and the static compensation capacity of the dynamic compensation nod in the corresponding group can be confirmed. The obtained data and the compensation capacity of the corresponding nod can be taken as the sample in the later dynamic reactive power optimization. Dynamic reactive power optimization utilized improved BP algorithm to build neural network, and carried on the net training and net test in the sample space. The trained BP network is obtained. This network has reached a millisecond level operation speed, so it can ensure the compensation capacity of dynamic reactive optimization.Finally, taking the actual 6kv distribution line as an example and combining the dynamic stimulation method, the paper has calculated and analyzed the proposed genetic algorithm of nerve network, and compared it with other algorithms. The result demonstrates the high-efficiency and practical applicability of this algorithm when it is applied in the reactive power optimization of oilfield network.